Biocomputing

Gibson Group

EMBL

Multiple Alignment and Phylogenetic Tree Practical

by Toby Gibson

for the EMBO DNA Sequencing Course 11-21 / 11 / 97

The Aims of this practical are 2-fold:

• 1. Utilise the mitochondrial DNA sequences that you generated earlier in the course in a phylogenetic comparison with a reference set of human polymorphisms.
• 2. Familiarise yourself with some freely available software:
  • Clustal X for doing alignments and Neighbour-Joining trees
  • NJplot a simple program for displaying trees.

Background

The mitochondrial D-loop or control region is probably the fastest mutating sequence region in animal DNA. It is therefore very useful for comparing closely related organisms (but useless for comparing distantly related species). Polymorphisms in the D-loop can be used for intra-species comparisons. In recent years, mtDNA polymorphisms have become popular for tracing the history of human migrations in the time span -100,000 years to present. The European populations are found to be rather homogeneous and it appears that a substantial migration of Indo-European speakers spread across Europe bringing farming technologies with them. A few European populations such as the Basques and Fins (who do not speak Indo-European languages) appear to be distinguishable from the farmers, but all Europeans are still closely grouped. The most consistently divergent groups of people, as judged by frequencies of polymorphisms, are in Africa. This is consistent with the origin of modern humans in Africa and a founder effect in all other present day humans: migration of some smallish populations out of Africa was followed by a large expansion in population size of these non-African groups.

D-loop Reference Alignment

We have taken a selection of 20 D-loop sequences representing diverse sets of humans, from the collection of Vigilant et al. (1991) Science, 253, p. 1503. The sequences mostly have a large (artificial) deletion between the two most mutable segments of the D-loop. Trees made from this data set show the greater divergence of the Africans and the much tighter grouping of European and Asian sequences. Since most of the students on the course are of European origin, we would expect the new sequences to group closely with the reference Europeans. It will be interesting to see if there are any exceptions to this....

Materials for the practical

Each PC is provided with a directory containing:

• Clustal X- a program for multiple alignments and calculating trees by NJ.
• NJplot - for displaying trees calculated by Clustal X or other packages
• mito_set.aln - the prealigned reference set of human mitochondrial sequences.

In addition, you need your own sequence in FASTA format (or another format compatible with Clustal X).

Part 1. Aligning the new sequence to the reference alignment by Profile Alignment

Clustal X is a widely used, portable alignment program that allows the user to approach alignment in a flexible manner. Clustal X allows you to reuse an old alignment and add new sequences to it, or even merge two alignments together. In Clustal X, this is known as profile alignment. This is useful in any on-going project where new sequences are being generated and alignments need updating. Adding new sequences to an old alignment has some advantages. Firstly it is a lot faster than redoing the alignment from scratch each time. Secondly, the original sequence alignment is kept intact - especially useful if the alignment had needed hand editing.

Profile Alignment in Clustal X:

• Start Clustal X by clicking on the icon and a new window should open.
• Familiarise yourself with the layout and menus.
  • (Only on PCs: Unfortunately the Clustal X window size cannot currently be changed.)
• Toggle from Multiple Alignment Mode to Profile Alignment Mode. A second window appears for sequences.
• Select Load Profile 1 and read in the mitochondrial DNA alignment file (mito-set.aln).
• Use the sliders to survey the alignment - note the large artifical gap in many sequences.
  • This gap is in a region that is not informative as it does not mutate at high frequency.
• Now select Load Profile 2 and read in your own sequence.
• Now select Align Sequences to Profile 1.
• When the alignment is complete, toggle back to Multiple Alignment Mode and the new sequence will merge to the old dataset.
• Using the sliders review the alignment.

Questions

• Correct alignment is by no means guaranteed! Are there any obvious errors in the sequence alignment?
  • If so are there many ambiguities in your incoming sequence?
• Does your sequence span the complete reference alignment, or only a bit of it?
  • This could have a bearing on the phylogenetic information and tree quality.
• Do you find coloured alignments useful?
  • Do you like the colour scheme? If not, why not? Note that you can customise it yourself!

Part 2. Calculating a tree in Clustal X and displaying it with NJplot

Clustal X can calculate a basic tree using the Neighbour-Joining method, but it cannot display it. There are a number of tree display packages available. In the Clustal X package, we distribute a simple but useful (and portable) tree display program NJplot, provided by Manolo Gouy (Lyon). NJplot displays the tree as a dendrogram and allows you to do some basic manipulations, most importantly re-rooting the tree. For more serious phylogenetic work, Manolo Gouy also provides the PhyloWin/SeaView package, which can do much more than we cover in the practical today. There are of course many other tree packages.

Calculating a tree with Clustal X:

• Look in the Trees menu and try to understand the options.
  • Bootstrapping is a resampling method to check the stability of the nodes in the tree.
• Toggle on Excluding positions with gaps.
  • Question: Why is this likely to improve the tree?
• Select Draw N-J Tree - a new window pops up.
• Note the tree file name in the new window and click on OK.
• The tree file is a set of nested brackets that is hard for a human to read, but easy for a computer.
• You could also make trees with other permutations:
  • A bootstrapped tree might be informative.
  • But correcting for multiple substitutions would not make a difference as these sequences are so similar, there aren't going to be any...

Displaying the tree with NJplot:

• Click on the NJplot icon and a new window should open.
• Familiarise yourself with the layout.
• Click on the Open button and the file input window will appear.
• Give the tree file name and click on OK. The tree will be displayed.
• Find the Chimp sequences: they are the outgroup. Are they correctly rooting the human sequences?
• If not, click New outgroup, and then on the # character marking the Chimp's node.
  • Play around with the re-rooting if you like. See how silly some trees look.

Questions

• Horizontal branch lengths are proportional to divergence. Which human groups have the most divergent polymorphisms?
  • Is this consistent with the "Out of Africa" hypothesis?
• Where has your new sequence rooted?
  • With the Europeans?
  • With another group of humans?
  • On the branch leading to the last common ancestor shared with Chimpanzees?
• If the tree topology seems rather surprising, why do you think this might be so?

Finally we will print the trees (probably using another program TreeView), so that we can compare and discuss the results. See one of the demonstrators to get the printing done.